FrontierMath, tiers 1 to 3 leaderboard
- GPT-6.1 Sol leads
- 93.7%
- models measured
- 79
- at the top
- Tied
- median
- 57.5%
Leader on FrontierMath, tiers 1 to 3 · read October 21
On October 21, 2026, 79 of our catalog's models had a FrontierMath, tiers 1 to 3 result, and OpenAI GPT-6 Astra led with 93.7%.
Unpublished research-level math problems written by mathematicians, run by Epoch AI. Share solved.
Role in the index
We collect this benchmark and show it as supporting evidence. It is not part of the Vibe Coding Index, because it measures something other than coding and the index is about coding. Read the methodology
| # | Model | Result | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|
| 1 | GPT-6 AstraOpenAI | 93.7% | 88.7 | $50.00 | |
| 2 | GPT-6.1 SolOpenAI | 93.7% | 88.7 | $10.00 | |
| 3 | Claude Opus 5.5Anthropic | 91.2% | 92.2 | $20.00 | |
| 4 | Claude Fable 5.1Anthropic | 90.2% | 89.1 | $50.00 | |
| 5 | GPT-6 SolOpenAI | 89.8% | 82.8 | $10.00 | |
| 6 | GPT-5.6 SolOpenAI | 89.1% | 76.7 | $20.00 | |
| 7 | Claude Sonnet 5.5Anthropic | 88.8% | 88.9 | $10.00 | |
| 8 | GPT-5.5 ProOpenAI | 87.7% | Not benchmarked | $180.00 | |
| 9 | Claude Fable 5Anthropic | 87.0% | 82.1 | $50.00 | |
| 10 | Gemini 4 ArgonGoogle | 86.0% | 80.5 | $10.00 | |
| 11 | GPT-5.6 TerraOpenAI | 86.0% | 69.0 | $12.00 | |
| 12 | Claude Opus 5Anthropic | 85.6% | 81.8 | $25.00 | |
| 13 | GPT-5.5OpenAI | 85.3% | 67.8 | $30.00 | |
| 14 | GPT-5.4 ProOpenAI | 82.5% | Not benchmarked | $180.00 | |
| 15 | GPT-5.6 LunaOpenAI | 82.1% | 61.5 | $1.20 | |
| 16 | Claude Opus 4.8Anthropic | 80.0% | 66.0 | $25.00 | |
| 17 | GPT-6 LunaOpenAI | 78.9% | 71.4 | $0.50 | |
| 18 | GPT-5.4OpenAI | 78.6% | 37.7 | $15.00 | |
| 19 | Claude Haiku 5.5Anthropic | 75.1% | 75.9 | $0.50 | |
| 20 | Qwen3.8 Max (0902)Qwen | 74.7% | 63.5 | $6.00 | |
| 21 | Muse Spark 1.3Meta | 74.4% | 79.7 | $4.25 | |
| 22 | GPT-5.2 ProOpenAI | 74.0% | Not benchmarked | $168.00 | |
| 23 | Kimi K3Moonshot AI | 72.2% | 61.8 | $11.936 | |
| 24 | Gemini 3.7 FlashGoogle | 71.6% | 69.9 | $3.75 | |
| 25 | Claude Opus 4.7Anthropic | 70.2% | 51.1 | $25.00 | |
| 26 | GLM 5.3Z.ai | 68.8% | 73.1 | $1.40 | |
| 27 | Gemini 3.8 FlashGoogle | 68.4% | 70.5 | $4.125 | |
| 28 | GPT-5.2OpenAI | 67.4% | 58.9 | $14.00 | |
| 29 | DeepSeek V4.1 FlashDeepSeek | 67.4% | 77.2 | $0.525 | |
| 30 | Claude Opus 4.6Anthropic | 66.0% | 66.9 | $25.00 | |
| 31 | Grok 4.6xAI | 66.0% | 76.4 | $6.00 | |
| 32 | Claude Sonnet 5Anthropic | 65.6% | 63.5 | $10.00 | |
| 33 | DeepSeek V4 Pro 0813DeepSeek | 64.6% | 54.6 | $1.98 | |
| 34 | Qwen3.7 MaxQwen | 64.6% | 45.1 | $4.425 | |
| 35 | Gemini 3.5 FlashGoogle | 62.8% | 42.8 | $9.00 | |
| 36 | Gemini 3.1 Pro PreviewGoogle | 59.6% | 35.3 | $12.00 | |
| 37 | GLM 5.2Z.ai | 59.2% | 57.7 | $7.00 | |
| 38 | Gemini 3.6 FlashGoogle | 58.9% | 48.3 | $3.75 | |
| 39 | Mistral Large 4Mistral AI | 58.6% | 55.3 | $4.18 | |
| 40 | DeepSeek V4 Flash 0731DeepSeek | 57.5% | Not benchmarked | $1.32 | |
| 41 | Kimi K2.6Moonshot AI | 57.2% | 44.2 | $3.41 | |
| 42 | Grok 4.5xAI | 57.2% | 75.5 | $6.00 | |
| 43 | GPT-5 ProOpenAI | 55.8% | Not benchmarked | $120.00 | |
| 44 | GLM 5.3 FlashZ.ai | 55.8% | 71.0 | $0.50 | |
| 45 | GPT-5OpenAI | 55.4% | 72.2 | $10.00 | |
| 46 | Kimi K2.7 CodeMoonshot AI | 54.0% | 39.1 | $3.35 | |
| 47 | Grok 4.7xAI | 53.0% | 75.0 | $6.00 | |
| 48 | Gemini 3 Flash PreviewGoogle | 51.2% | 75.8 | $3.00 | |
| 49 | GPT-5.4 MiniOpenAI | 51.2% | 28.2 | $4.50 | |
| 50 | GPT-5 MiniOpenAI | 46.7% | 59.8 | $2.00 | |
| 51 | T Inkling SmallThinking Machines | 46.3% | 29.9 | $1.20 | |
| 52 | DeepSeek V4 Pro 0423DeepSeek | 45.3% | 35.3 | $1.9105 | |
| 53 | GPT-5.4 NanoOpenAI | 44.9% | Not benchmarked | $1.25 | |
| 54 | Grok 4.3xAI | 42.8% | 22.5 | $2.50 | |
| 55 | Qwen3.6 PlusQwen | 38.2% | 37.4 | $1.95 | |
| 56 | GLM 5.1Z.ai | 36.8% | 44.0 | $4.40 | |
| 57 | o4 MiniOpenAI | 36.1% | 68.6 | $4.40 | |
| 58 | Qwen3.6 27BQwen | 35.1% | Not benchmarked | $2.00 | |
| 59 | Claude Opus 4.5Anthropic | 34.4% | 66.8 | $25.00 | |
| 60 | Qwen3.7 PlusQwen | 34.4% | Not benchmarked | $1.28 | |
| 61 | o3OpenAI | 33.3% | 70.7 | $8.00 | |
| 62 | T InklingThinking Machines | 33.3% | 49.7 | $4.05 | |
| 63 | Qwen3.5 397B A17BQwen | 31.2% | 28.6 | $2.34 | |
| 64 | Gemini 3.1 Flash LiteGoogle | 27.7% | Not benchmarked | $1.50 | |
| 65 | Gemini 3.5 Flash LiteGoogle | 26.0% | 33.6 | $2.50 | |
| 66 | Gemini 2.5 ProGoogle | 24.6% | 37.0 | $10.00 | |
| 67 | Claude Sonnet 4.5Anthropic | 23.9% | 59.1 | $15.00 | |
| 68 | Qwen3.6 FlashQwen | 22.5% | Not benchmarked | $1.125 | |
| 69 | Qwen3.6 35B A3BQwen | 20.4% | Not benchmarked | $0.95 | |
| 70 | GPT-5 NanoOpenAI | 20.0% | 34.8 | $0.40 | |
| 71 | Qwen3.7 FlashQwen | 19.3% | Not benchmarked | $0.13 | |
| 72 | o3 MiniOpenAI | 18.6% | 52.1 | $4.40 | |
| 73 | Qwen3.5-FlashQwen | 18.2% | 6.2 | $0.26 | |
| 74 | o1OpenAI | 14.7% | 61.7 | $60.00 | |
| 75 | Claude Opus 4.1Anthropic | 12.6% | 27.2 | $75.00 | |
| 76 | GPT-4.1 MiniOpenAI | 6.7% | 28.5 | $1.60 | |
| 77 | GPT-4.1OpenAI | 6.0% | 46.5 | $8.00 | |
| 78 | GPT-4o-mini (2024-07-18)OpenAI | 0.7% | 3.6 | $0.60 | |
| 79 | GPT-4o (2024-08-06)OpenAI | 0.4% | 23.1 | $10.00 |
Only models in our catalog are listed, and each appears once with its best published run. 197 rows were read from the source on October 21, 2026, and rows that match no catalog model are left out.